Research On Tracking Control Algorithm Of Underwater Vehicle Based On Reinforcement Learning

As the rich resources in the ocean are gradually recognized,the ocean has become an important place for countries to strive for ocean rights and interests,develop cutting-edge technology,and display comprehensive national strength.Autonomous underwater vehicle is an important tool for exploring ocean and carrying out ocean mission.With the increasing demand for marine exploration,it is urgent to improve the autonomy of underwater vehicles and the reliability of risk events.The trajectory tracking technology of autonomous underwater vehicle is an important key technology in the control of underwater vehicle.The accurate tracking of trajectory is the basis of the task of autonomous underwater vehicle in complex marine environment.At present,there is little joint research on the tracking and positioning of autonomous underwater vehicle.In this paper,considering the uncertain constraints of underwater vehicle model,the trajectory tracking and formation control algorithm of autonomous underwater vehicle are studied.The main work of this paper is as follows:(1)For the problem that the error of autonomous underwater vehicle in trajectory tracking is large and the control strategy is difficult to adjust,a tracking control method based on reinforcement learning is designed.In order to reduce the calculation of system mean,a multivariate probability configuration method is used to sample the uncertain parameters of autonomous underwater vehicle.Finally,the simulation shows that the proposed algorithm can achieve good track tracking effect.(2)For the problem of autonomous underwater vehicle position uncertainty under the constraint of dynamic model,the joint design of positioning and tracking control is carried out.On the basis of accurate acquisition of autonomous underwater vehicle position,a tracking control algorithm based on reinforcement learning is designed.The tracking control accuracy of underwater vehicle is improved and the tracking error is reduced.Finally,the effectiveness of the trajectory tracking algorithm is verified by simulation and pool experiment.(3)On the basis of the tracking and control problem of single autonomous underwater vehicle,the formation control problem of multiple autonomous underwater vehicles is further considered.Aiming at the coordination problem of multiple autonomous underwater vehicles tracking,a directed topology diagram of autonomous underwater vehicle formation is established.A formation control method based on reinforcement learning is designed to make up for the deficiency of single autonomous underwater vehicle tracking.